Reading new business formations by ZIP code
Counting new business formations by ZIP code looks like the simplest analysis in registry data: group filings by ZIP, sort, done. Run it on raw Florida data and the top ZIP is likely to be one that houses registered-agent offices, not one where new businesses are opening. Four corrections turn a misleading count into a usable map.
Key takeaways
- Exclude agent addresses (1,000+ entities at one street address) before ranking ZIPs, or you rank filing services.
- End monthly trends at the last complete month; the current month always looks like a collapse.
- Normalize by residents and by employer sites, and read the two together.
- Treat "share still active" for the last two formation years as an upper bound.
Correction 1: Strip out agent addresses
A registered-agent office or virtual-mailbox provider can carry tens of thousands of entities at one street address. Put that address in a ZIP and the ZIP tops every formation chart. CoverFi's explore views define an agent address as any address with 1,000 or more registry entities, unit numbers merged, and exclude those entities by default from the formations and density views.
The shared-addresses view is the other side of the same coin: it ranks street addresses by how many entities they carry, with a last-12-months count beside the all-time total. A sudden jump in new registrations at one address usually means a new filing service, and sometimes means something worth a closer look.
Correction 2: Use complete months only
States post filings 1 to 5 business days after they are filed, and late filings keep arriving for weeks. A chart that includes the current month will show a cliff at the right edge every single time. The fix:
- End the 12- and 24-month windows at the last complete calendar month.
- Compare year-to-date against the same day last year, not against last year's full total.
- Expect the most recent complete month to be revised upward slightly as late filings land.
Correction 3: Normalize, and use two denominators
A dense urban ZIP will out-form a rural one on raw counts regardless of business climate. Normalizing fixes that, but each denominator has its own bias:
| Measure | Denominator | Overstates | Use it for |
|---|---|---|---|
| Entities per 1,000 residents | Census ACS population | Business districts with few residents | Residential and suburban areas |
| Entities per employer site | Census ZIP Business Patterns establishments | Areas full of home-based or paper entities | Spotting registrations without operations |
| Raw entities | None | Big ZIPs | Sizing a territory |
Read the first two side by side. A ZIP high on entities per resident but ordinary on entities per employer site is a business district. A ZIP high on both is genuinely registration-heavy. Filter out tiny ZIPs before ranking per-resident ratios; CoverFi's density view defaults to ZIPs with at least 10,000 residents for this reason.
Correction 4: Read survival with the reporting cycle in mind
The lifecycle view groups entities by formation year and shows how many from each year are still active. It is a good proxy for churn, with one trap: an entity is usually marked inactive only after it misses an annual report. Companies formed in the two most recent years have not been through a full annual-report cycle, so their "still active" share is an upper bound, not a survival rate.
What the formation data covers
CoverFi's explore views draw on the state registries in its business dataset: Florida, New York, Connecticut, Colorado, Texas, Pennsylvania, and Alaska. A few details shape what you can conclude:
- The state filter means the principal address state, not the formation state.
- Entity mix splits LLCs, corporations, and other types, which is a useful signal: a ZIP where nearly all new filings are LLCs is mostly small operators and holding companies.
- Population and establishment counts come from Census ACS and ZIP Business Patterns, so they lag the filings by a year or more.
For the fields behind each filing and how states differ, see the business entity search guide. For the area statistics themselves, CoverFi's public ZIP Code profiles show income, housing, and business mix for each ZIP with county, state, and national percentiles.
Frequently asked questions
Why does one ZIP have far more new LLCs than any other?
Usually because a registered-agent office or virtual mailbox sits in it. Exclude high-volume agent addresses and the ranking changes.
Is the Census Business Formation Statistics series the same thing?
No. The Census series counts applications for employer identification numbers and is published mainly at national and state level. Registry formations count entities a state actually created, and can be cut by ZIP.
Why did formations drop in the latest month?
Almost always a partial month. Filings post with a delay, so trend charts should stop at the last complete month.